Abstract This paper describes an approach to smoothing and categorizing noisy data, using neural networks. The specific example is that of interpreting product order data as a preliminary step to forecasting future orders. However, the technique is quite general and can be applied to noisy data in a wide variety of systems, including natural, social and artificial ones. In recent years, a major characteristic of computer and electronic products has been their short life cycles. This creates unprecedented challenges, particularly for the task of sales volume forecasting in manufacturing enterprises. Owing to the short history of available past order data, traditional forecasting methods are not adequate. In this paper, a robust back-propagation neural network is adopted to extract an underlying shape for the order pattern of each product. The self-organizing map neural network is used to categorize products according to these shapes. The categorization results will establish a basis for further investigation toward new production introduction analysis.
[1]
Geoffrey E. Hinton,et al.
The appeal of parallel distributed processing
,
1986
.
[2]
Teuvo Kohonen,et al.
Self-Organizing Maps
,
2010
.
[3]
David M. Skapura,et al.
Neural networks - algorithms, applications, and programming techniques
,
1991,
Computation and neural systems series.
[4]
Ramesh C. Jain,et al.
A robust backpropagation learning algorithm for function approximation
,
1994,
IEEE Trans. Neural Networks.
[5]
Robert Tibshirani,et al.
Bootstrap Methods for Standard Errors, Confidence Intervals, and Other Measures of Statistical Accuracy
,
1986
.
[6]
Takayuki Ito,et al.
Neocognitron: A neural network model for a mechanism of visual pattern recognition
,
1983,
IEEE Transactions on Systems, Man, and Cybernetics.
[7]
James L. McClelland,et al.
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations
,
1986
.